Senior AI Security Engineer

What You'll Bring • 3–5+ years in security engineering or DevSecOps • Foundational understanding of ML concepts, workflows, and frameworks (PyTorch, TensorFlow, scikit-learn) • Working knowledge of LLMs, transformer architectures, and their security concerns (prompt injection, jailbreaking, data poisoning, model extraction) • Container and Kubernetes security experience, including securing AI workloads • Hands-on Python plus Bash or Go, and comfort with common data science libraries • Cloud security in AI contexts — GPU security, distributed training, and data protection in ML pipelines • Infrastructure-as-code experience (Terraform, Ansible) • Solid grounding in encryption, key management, and PII/data privacy • Familiarity with secure SDLC, ML supply-chain security, and model governance/versioning • Working knowledge of compliance frameworks as applied to AI (GDPR, SOC 2) • Strong communication skills and the ability to work independently or on a team

What You'll Do • Assess the security posture of LLMs and generative AI systems used within or by Applied Systems • Run threat modeling and security architecture reviews for AI/ML systems and their data pipelines • Implement and maintain security controls across model training, fine-tuning, and inference infrastructure • Identify and help remediate AI-specific vulnerabilities — prompt injection, model poisoning, data exfiltration, and adversarial attacks • Build data security and privacy controls for training datasets and inference inputs • Develop security baselines, hardening configs, and incident-response runbooks for AI platforms • Implement monitoring and detection for anomalous AI system behavior • Contribute to internal AI security policies, vendor reviews, and security training

Nice to Have • Security certification (Security+, CISSP) • Hands-on experience with OpenAI / Anthropic Claude / Vertex AI / Azure OpenAI • ML security tools (Robust Intelligence, Arthur AI) • AI security research or publications • Penetration testing • Red-teaming AI systems